Association Analysis of Wind Turbine Grid-Connected Oscillation Modes and Influencing Factors Based on Improved Association Rule Mining

Accurate oscillation mode recognition and stability analysis based on big data are critical for the safe operation of wind turbine systems. This paper utilizes modern statistical and machine learning methodology to analyze the correlation between monitored wind turbine operation data and oscillation phenomena, and a system oscillation analysis and diagnosis method is proposed based on an improved association rule mining (ARM) model. Firstly, the oscillation modes in the power data are measured by the synchronous extraction transform. By improving the ARM model, a thorough study is conducted on the correlation between oscillation modes and variables such as wind speed, compensation degree, voltage fluctuation, etc. Finally, the component importance measure is used to optimize each element's risk weight calculation method relative to the system oscillation. The experiments demonstrate that the proposed association rule analysis method can effectively analyze the relationship between system oscillation phenomena and influencing factors and exhibits high diagnostic accuracy.